JavaLand 2024: Application Development Green Masterplan
•
0 likes•125 views
My presentation slides I used at JavaLand 2024
More Related Content
Similar to JavaLand 2024: Application Development Green Masterplan
Similar to JavaLand 2024: Application Development Green Masterplan (20)
More from Miro Wengner
More from Miro Wengner (12)
Recently uploaded
Recently uploaded (20)
JavaLand 2024: Application Development Green Masterplan
- 1. Source: marvel In 2008 Tony Stark introduced… ARC fusion reactor Affordable Robust Compact MIT continues research
- 3. Miroslav Wengner Green Masterplan Application development
- 4. Miroslav Wengner • Husband, Father, Software Engineer, Author, Blogger, Technology Enthusiast • Book Author: Practical Design Patterns for Java Developers [Packt] • OpenJDK Committer, OpenSource projects contributor, Ex JCP Executive Committee • Co-Author of Robo4J Project (Duke Award) • Java Champion, JavaOne RockStar • Principal Engineer at OpenValue
- 5. Safe Harbor Statement All what you will hear can be di ff erent, this presentation is for motivational purposes …
- 6. Agenda • History and Evolution • Industrial Revolution • Green Initiatives and Energy • Energy Consumption in action • Conclusion • Q/A
- 7. Mechanical The ability to make tools became commonplace Electrical Source: https://en.wikipedia.org/
- 8. Mechanical may have a Similar Dominators Electrical Source: https://en.wikipedia.org/ e ff ective resources usage per unit
- 9. Data Centers Software How did we got there ?
- 10. Industrial Revolution: timeline ? Source: https://en.wikipedia.org/ Phase 1. Steam Engines Factories Textile Machinery small spot 18th 19th 20th 21st Phase 2. Logical Systems Electri fi cation Assembly lines Bigger Scale expanding spots Phase 3. Transistor Optical Ampli fi er Computers Internet Digital information at the large scale Internet of Things Cloud Computing Arti fi cial Intelligence Machine to Machine communication connecting spots
- 12. Industrial Revolution ? Source: https://en.wikipedia.org/ Sea level has risen about 250 mm Annual Surface temperature has risen about 1,5 Celsius …Past is knocking
- 13. Industrial Revolution: connected spots Laptops, Desktops, IoT Devices, Cars, Smart Houses, AI Servers, Virtual Machines, Clouds Provides Energy resources for SOFTWARE ware taken as granted ? taken out of the equation ? SEEMS POSITIVE
- 14. Energy: Data Centers • Globally, by 2025, they are projected to consume 20% of global electricity and emit up to 5.5% of the world's carbon emissions. • signi fi cant part of the energy consumed is transformed into: • heat may reduce a system reliability • life expectancy of devices • escalation in cooling requirements • Virtual Machines may run Idle (maybe 40->80%)
- 15. Initiatives: Green Software Foundation Software Carbon Intensity E: Energy consumed by software in kWh I: Carbon emitted by kWh of energy, M: Carbon emitted through the hardware R: Scaling factor what about the real emissions ?
- 16. Energy: Software industry impact Scope 1+2 : own operations Scope 3: broader value chain contains customers, suppliers
- 17. Green Initiatives: take aways 1. Carbon E ffi ciency implies cost reductions (not surprise) 2. Green Software Development helps introduce coding best practices (?) di ff erent understanding business vs. nature 3. Shape demand for application development - better forecasting 4. Reduce carbon emissions 5. Increase lifecycle and usefulness of hardware and servers Sustainability
- 18. Green Initiatives: take aways 1. Software Industry should take a RESPONSIBILITY and accept challenges Only One Planet to Sustain Energy as a good PROXY to emissions footprints
- 19. Energy Consumption: how to … 1. precisely de fi ne a task constrains 2. select tools 3. fairly measure 4. design process 5. undestand a data Task Watt Measurement Time t_x Performed action Execution Time T_ex Energy_Time T_en Driving Idea
- 20. Energy Consumption: selecting tools ? for i in range(range_max): if is_prime(i) : found_primes += 1 Python for (int i = 2; i < range_max; i++) { if(is_prime(i)){ found_primes++; }; } C for (var i = 0; i < range_max; i++) { if(isPrime(i)){ foundPrimes++; } } JavaScript for (int j = interval.start; j < interval.stop ; j++) { if (isPrime(j)) { foundPrimes.incrementAndGet(); } } Java How to Repeat ? Execution TOOLS: • shell script - PID • Tinytuya framework
- 21. ? … AUTOMATION … Energy Consumption: fair measurements
- 22. ? Energy Consumption: measure fairly
- 23. Energy Consumption: measure fairly Simple ideas could become complicated … AUTOMATION… rock it ! Measure IDLE state to normalize results
- 24. Energy Consumption: action Selected Platforms 1. Java: OpenJDK, Zulu ver17, 21 2. C: gcc 12 3. Node 20 4. Python 3.11 Hardware 1. Raspberry PI 4 : 4c 2. Linux Intel i7 : 4c 3. MBP i7 : 4c and 6c 4. MBP : M1 Max, M2 Max Calculated: Mean Value (MV), Standard Deviation (SD), Standard Error of Mean (SEM)
- 25. Energy Consumption: data ? Task: Find Prime Numbers Range: from 0 to 200 000 000, repeat 15 times Watt Measurement: Idle State, Automatically calculated time window ( probes ) Java code is design with minimal heap usage Task Duration varies Some languages show signi fi cant energy consumption increment
- 27. Energy Consumption: kWatt/Task ? Energy Consumption Standard Error of Mean (SEM) = 2,95 %
- 28. Energy Consumption: understanding data ? MacBookPro 2019, Intel i7, 6 cores MacBookPro M2 Max best execution cca. 5 times slower cca. 24 times lower Energy consumption
- 29. Energy Consumption: understanding data ? Di ff erence between ARM and Intel architecture
- 30. Conclusion ? • Who: Software Companies • Managers, Devs, Ops • What: should take a responsibility • Why: CO2 footprint
- 31. Conclusion ? • contribution to Green Masterplan: • platform selection • application design • each executed instruction may contributes to CO2 emission • extended tooling • coding style and patterns Book Author: Practical Design Patterns for Java Developers [Packt] • controlled, not overused parallelism matters • hardware may be good enough => usefulness
- 32. Thank YOU ! twitter: @miragemiko github:@mirage22 email: miro@openvalue.de
- 33. Resources • Energy-e ffi ciency and sustainability in new generation cloud computing: A vision and directions for integrated management of data centre resources and workloads : https://doi.org/10.1002/spe.3248 • TechMonitor: https://techmonitor.ai/focus/tech-industry-carbon-emissions- progress • Climate Change: Atmospheric Carbon Dioxide: https://www.climate.gov/ • Green Software Foundation: https://greensoftware.foundation/ • Energy E ffi ciency of Programming Languages on Arm and Intel: https:// openvalue.blog ?